Detection Engineering

Custom detection rules with AI-assisted tuning. Correlation engine maps alerts to MITRE ATT&CK techniques automatically.

Within the Security Operations zone, Detection Engineering represents a critical operational capability that DevOps AI delivers through its unified platform. This process area operates at HITL Gate Level L2 (Approve to Proceed), meaning AI prepares recommendations and stages actions, but a designated human must explicitly approve before execution proceeds.

DevOps AI implements Detection Engineering as a fully integrated workflow within the Security Operations zone. When deployed from the Azure Marketplace, this process area is automatically provisioned with role-appropriate dashboards, notification rules, and automation policies tailored to your MSP's operational requirements.

Workflow Architecture

The Detection Engineering workflow follows DevOps AI's standard event-driven architecture. Events are ingested through the platform's connector framework — pulling data from PSA tools (ConnectWise, Datto Autotask, HaloPSA), RMM platforms (NinjaRMM, Datto RMM), and Microsoft 365 tenants — then processed through the AI inference pipeline before reaching the L2 gate for human review.

Multi-Tenant Isolation

Every operation within Detection Engineering respects DevOps AI's zero-trust multi-tenant architecture. Client data is isolated at the Azure tenant level, encrypted at rest with customer-managed keys, and processed within geo-fenced compute boundaries. No cross-client data leakage is possible — even AI models are trained on anonymized, aggregated patterns rather than raw client data.

Detection Engineering is classified at HITL Gate Level L2, which defines exactly when AI acts autonomously and when human judgment is required. This classification was determined through risk analysis of the process area's blast radius, reversibility, and compliance implications.

Why L2?

This process area involves high-impact or partially-reversible actions where human judgment adds critical value. AI handles the analysis, preparation, and recommendation — but execution requires explicit human approval. This balances efficiency with risk management.

Detection Engineering does not exist in isolation — it integrates with other process areas across the Security Operations zone and the broader DevOps AI platform through the event mesh architecture. Actions in this process area can trigger workflows in related zones, and events from other zones can feed into Detection Engineering operations.

Connector Framework

DevOps AI's connector framework provides bi-directional integration with the tools MSPs already use. For Detection Engineering, this typically includes PSA platforms (ConnectWise Manage, Datto Autotask, HaloPSA), Microsoft Graph API (Azure AD, Intune, Defender), and specialized third-party tools relevant to Security Operations operations. All connectors are managed through the platform's Marketplace zone — install once, available everywhere.

Analytics & Reporting

Every operation within Detection Engineering generates structured telemetry that feeds into the Analytics zone. Dashboards provide real-time visibility into process area health, throughput, error rates, and HITL override frequency. Over time, the AI models learn from human overrides to improve future recommendations — creating a continuous improvement loop that makes Detection Engineering more accurate with every interaction.

Audit Trail

Complete audit provenance is maintained for every action within Detection Engineering. This includes the triggering event, AI analysis results, HITL gate decisions (including who approved and when), execution outcomes, and any rollback actions. Audit data is immutable, tamper-evident, and exportable in OSCAL format for compliance evidence collection.